Filter links can be applied to dataframes to filter the rows of a dataframe down to a specific subset. In the example below, we use two charts to generate filters, one that only selects customers that are employed, and one that selects customers who are married. These two filters are then ANDed together to filter the input dataframe of the table operator (note this is configurable to support the full range of boolean operations: AND, OR, NOT).
Note that filter links can be ORed, by clicking on the icons where the filter links meet. When the icons are vertically arranged, they're being ANDed. When they are horizontally arranged, they're being ORed.
Filter links can also be inverted, by clicking on the link itself:
When there's a filter link from one chart to another, an additional option is to create a brush filter. This is indicated by the link's color changing, and means that the filter is used to show different subsets of the data.
The semantic difference to dataframe links is that instead of flowing dataframes to an operator, filter links flow predicates. So rather than outputting a new dataframe defining the structure of columns, they output a filter expression that can be applied to a dataframe (e.g.,
Employed = true).